Speed Test Accuracy

Minoman

I have a Fingbox running Fing Service 2.2.0 on an iMac running MacOS 10.15.6 (Catalina)

When I run the Internet speed test within my Fing Client, it tells me that my download speed is 2.9 MBPS and my upload speed is 0.4 MBPS 

If I then run the industry standard test, Speedtest by Ookla, it tells me that my download speed is 34.1 MBPS and my upload speed is 05.45 MBPS

Is there any explanation for this massive discrepancy?  

Marc

@Minoman, I believe the Fing desktop uses mlabs as its testing provider.  Could you run a test from their website and then from Fing’s desktop and see if the results match?  https://speed.measurementlab.net/#/

If your see a discrepancy using the above link, then getting Fing involved would be in order.

Minoman

Hi Marc,
I've done as you've suggested and it has raised more questions than answers
M-Labs = 4.98 down, 2.18 up, 113mS latency
Fing = 7.5 down, 3.7 up, 37.1mS jitter
Ookla Speedtest = 27.9 down, 4.59 up, 87mS jitter
BT-Wholesale Speed Test = 34.98 down, 6.81 up
So who do you believe?
I'd like to think that it's the Fingbox as it is a dedicated hardware based analyser but I may be mistaken
So where might I go from here?
Mike

Marc

Thanks for testing.  What speed do you pay for so we have a reference.  Also, on m-labs page, there is a support email to raise an issue, [email protected]  As your seeing this issue from their web page as well as Fing, it might be worth raising a support query with them.

@Robin, do you have any suggestions?  The differences between m-labs and ookla seem pretty significant.  

BK303

Fingbox can run an internet speed test on a schedule and on demand, the Fing App on a smartphone can run a speed test direct from phone to the internet or a WiFi speed test between the phone and the local Fingbox or trigger the Fingbox to start a test, the Fing client on a Mac or PC apparently can run a speed test too but unclear if it's triggering the Fingbox or going direct to the internet based on the image shared (I don't use Mac Client)....so there are a lot of tests available and we need to be clear which is being used while troubleshooting. When testing in a browser to Ookla or ISP, the Fingbox is not involved.

Likely the highest results you are seeing when testing in browser do represent what your network is capable of achieving. For most consistent results, I suggest remove all WiFi and only connect by Ethernet wire into your router or modem. If your Fingbox tests are consistently low compared to other tests (ideally also using mLabs as already suggested), try using a different Ethernet cable or different physical port on your router. Ethernet wired link could be operating at 10,100, or 1000 Mbps mode depending on equipment and cable. To check the operating speed mode of the Fingbox wired connection, from Fing phone app select the fingbox network and then gear icon in upper-right, which will show version of Fingbox software and MAC..scroll down to "Info" and then look at "link Speed". Alternatively, look at router status lights for the Ethernet port or router management page to determine what speed the Fingbox link is operating.

Good luck.

~BK

Minoman

To answer Marc's question, my Broadband contract with BT is for a minimum 20 MBPS
The regular Fing reports put BT in the bottom 25% of providers which I always thought a little strange when I was clocking around 30MBPS but then I was using Ookla whilst Fing was using >>>>

As for BK's comments
I am using a BT Smarthub 2 router and both my iMac and my Fingbox are ethernet linked to my Router on a iGB ethernet link
I have just run the Fing speed test on my iPhone and it shows a 3.9 down, 1.0 up speed
The Fing App on my iMac, a few minutes later, shows 2.9 down and 3.9 up
It's almost as if all is being done with random number generators!

Marc

I have absolutely seen speed differences between mlabs/ Fing / Ookla etc and what I pay for but not to the extent your seeing. Usually see a 10mbps or so difference between them.

It’s consistent so I kind of just accept it and made it my baseline. Yours is a bit much though.

BK303

I find it is not unusual for speed tests to vary +/- 30%, especially during the busiest parts of the day on my current cable modem service.  

From the data reported, I suspect that mLabs does not have a very good path from their nearest server to your local BT network.  The latency in your screen shot (545ms from Dublin) is far higher than most nearby servers used for speed testing, as most test services have many servers and the nearest one is automatically chosen.  My tests using Fingbox are typically 50 to 100ms latency.    To check latency to some servers, you can open a terminal window and experiment with ping commands to some hosts like "speedtest.net" (Ookla),  speed.measurementlab.net (mLab),  1.1.1.1 (Cloudflare global DNS), or 8.8.8.8 (Google global DNS).  

Unfortunately, since it seems you can't choose which server the Fingbox or FingApp are using, that doesn't leave you with many options for improving Fingbox speed test accuracy.    I would trust the numbers from Ookla and BT's test server to be good indication of the quality of the link from your ISP to your home router.  
Cheers,
BK

Minoman

Thanks BK - things have gotten busy - I'll do as you suggest and get back in a few days....

Lee_Bo

Fing speedtest: 225 down, 25 up
Mlabs speedtest: 205 down, 19 up
Speedtest: 235 down. 29 up.
Ubiquiti hardware, Spectrum 200 meg service.

CountZeroh

Basically, unless your fing box is directly connected to your internet you may get results that are 20-40% off the speed you pay for. So I pay for 400Mbit... It means that if your fing is behind a WiFi router I see 320 ish on a good day. And typically see 250-260Mbit.

Why is this:

1. Cable modems are in a shared ring topology. So...as an example... You share the connection on your street to your service provider. This means as more people go online, everyone gets slower.

2. Typical wifi routers people get. 50$-300$ ranges don't use powerful chips in it. This means as your home network gets busy... Your wifi router will be burdened which will cause internet slowness.

2b. Also remember that your wifi router is typically also your internet firewall. It's doing work to look at the packets you are sending out. That adds a little slowness as your home usage goes up.

3. There is such a thing as internet slowness. Internet test are basically a test from your device to another on the other side of the state or country. Anything in between can affect that.

3. Wired connections and tests will be much faster than wifi. Wired is typically now about 1 Gbit or 1000 Mbit. The fastest wifi is 480Mbit. Wifi is also subject to interference from microwaves.... Other Wifi... That wired connections are not.

Good luck, you may already know all this.

Minoman

Thanks CountZeroh
However both my FingBox and my iMac are ethernet connected to my Router with a 1GB link
So why is it that the FingBox tests, which I'd expect are firmware based, read significantly lower than the Ookla tests (App based) and the and the BTwholesale tests (web-based) when taken within minutes of each other? So no Wi-Fi effects and the state of the Internet is broadly the same for all the tests?

Robin_Ex_Fing

Hi @Minoman
Check this useful link which might help you: https://help.fing.com/knowledge-base/internet-speed/
Check the FAQ section also. Thanks

Lee_Bo

Minoman said:

Thanks CountZeroh
However both my FingBox and my iMac are ethernet connected to my Router with a 1GB link
So why is it that the FingBox tests, which I'd expect are firmware based, read significantly lower than the Ookla tests (App based) and the and the BTwholesale tests (web-based) when taken within minutes of each other? So no Wi-Fi effects and the state of the Internet is broadly the same for all the tests?

Well, a couple of things.
Both the Fingbox and Ookla are testing speeds from different destinations.  So let's say Fing tests from a destination on AT&T and Ookla tests from a spot on Windstream.  At your time of testing, Windstream might be having network issues that are affecting your testing, while AT&T is running just fine (yeah, right).  Then you may test and AT&T is running rough.  So circumstances outside your control can affect speed tests.
Also, your 1 GB link you mention is only for inside your home network, or from your router to your computers.  Anything past your router is at the mercy of any and all network congestion that exists in the world that we never see (unless you work in a NOC).